Modification of Plasma Membrane Protein Cysteine Residues by ADP-Ribose in Viuo*

Proteins can be post-translationally modified by ADP-ribose. Previously, two classes of ADP-ribosyl protein linkages have been detected in vivo which have chemical properties indistinguishable from ADP-ribosyl arginine and ADP-ribosyl glutamate or aspartate. Reported here is the detection of a third class of endogenous ADP-ribosyl protein linkage. This class is chemically indistinguishable from ADP-ribose linked to cysteine residues by a thioglycosidic bond. The distribution of ADP-ribosyl cysteine residues was studied in subcellular fractions of rat liver. Proteins modified on cysteine were detected only in the plasma membrane fraction. Pertussis toxin is known to disrupt signal transduction of ADP-ribosylation of cysteine residues of plasma membrane GTP binding proteins. The results described here raise the interesting possibility that the endogenous modification of plasma membrane protein cysteine residues may be involved in signal transduction.


Proteins
can be post-translationally modified by ADP-ribose.
Previously, two classes of ADP-ribosyl protein linkages have been detected in uiuo which have chemical properties indistinguishable from ADP-ribosyl arginine and ADP-ribosyl glutamate or aspartate. Reported here is the detection of a third class of endogenous ADP-ribosyl protein linkage. This class is chemically indistinguishable from ADP-ribose linked to cysteine residues by a thioglycosidic bond. The distribution of ADP-ribosyl cysteine residues was studied in subcellular fractions of rat liver. Proteins modified on cysteine were detected only in the plasma membrane fraction.
Pertussis toxin is known to disrupt signal transduction of ADP-ribosylation of cysteine residues of plasma membrane GTP binding proteins.
The results described here raise the interesting possibility that the endogenous modification of plasma membrane protein cysteine residues may be involved in signal transduction.
NAD' is a substrate for many dehydrogenases that catalyze hydride transfer reactions central to energy metabolism.
It is also the substrate for enzymes that catalyze the cleavage of the linkage between nicotinamide and ribose and the transfer of ADP-ribose to a nucleophilic acceptor. Such ADP-ribose transfer reactions represent a versatile mechanism for the post-translational modification of proteins. For example, poly(ADP-ribose) polymerase catalyzes transfer of ADP-ribose to protein carboxylate groups and to ribosyl hydroxyls of ADP-ribose resulting in the modification of proteins with ADP-ribose polymers (1,2). While all other known ADPribosyltransferases catalyze the transfer of only single ADPribose groups, they show a wide range of specificity for acceptors. The best understood mono-ADP-ribosyltransferases are the bacterial toxins which modify a specific amino acid residue in a specific target protein. in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. gation factor 2 (3). Cholera toxin and Escherichia coli heat labile enterotoxin modify a guanidino nitrogen of an arginine residue of G,, a stimulatory GTP binding protein involved in the regulation of adenylate cyclase (4,5), and pertussis toxin modifies the thiol group of a cysteine residue of Gi, an analogous inhibitory protein of the cyclase system (6-8). Clostridium botulinum C2 toxin modifies an arginine residue of nonmuscle actin (9, lo), and C. botulinum C3 toxin appears to modify an asparagine residue of a 21-kDa molecular mass membrane protein (11).
The synthesis of ADP-ribose polymers is known to occur at rapid rates in animal cells following DNA damage and is required for cellular recovery from DNA damage (1). However, evidence is accumulating that mono-ADP-ribose transfer reactions are also a versatile component of metabolism although the function is unknown.
Endogenous mono-ADP-ribosyltransferases include enzymes that catalyze the hydrolysis of NAD' to nicotinamide and free ADP-ribose (12) and enzymes which can modify proteins (13-16). Several apparently distinct transferases that can modify free arginine, other guanidino compounds, and arginine residues in proteins have been characterized (13), and an activity that modifies the diphthamide residue in elongation factor 2 has been reported (14,15). Most recently, an activity that modifies free cysteine and cysteine residues in proteins has been described (16). Evidence that the endogenous arginine-specific mono-ADP-ribosyltransferases actually modify protein in uiuo has come from the detection of proteins that are covalently modified with ADP-ribose by linkages indistinguishable from ADP-ribosyl arginine (17). Endogenous protein linkages with properties indistinguishable from ADP-ribosyl carboxylate ester have also been detected (17). Reported here is the discovery of a third class of protein ADP-ribose linkages in rat liver in vivo. These linkages are characteristic of modification at cysteine residues. Further, we report that proteins modified by ADP-ribose on cysteine are located exclusively in the plasma membrane fraction of rat liver.
The pellet was washed twice with 500 ~1 of 20%  (24), and a trichloroacetic acid-insoluble fraction was prepared and analyzed for ADP-ribosyl cysteine linkages as described above. Each fraction was also assayed for Na+,K+-ATPase activity as described by Quist et al. (23).

AND DISCUSSION
The use of neutral hydroxylamine to release ADP-ribose from carboxylate ester linkages on glutamate or aspartate and glycosylic linkages of the guanidinium group of arginine have been reported previously by this laboratory (17,ZO). Cysteine and diphthamide linkages to ADP-ribose are stable to hydroxylamine (17). It has been reported that thioglycosidic linkages can be cleaved in the presence of mercuric ion (251,and Meyer et al. (26) have recently reported that ADP-ribosyl cysteine linkages in transducin formed in vitro by the action of pertussis toxin can also be cleaved using this reagent. In order to search for proteins modified in uiuo by ADP-ribose  The indicated proteins radiolabeled with ADP-ribose were added to crude extracts of rat liver protein and analyzed for release by mercuric ion as described under "Experimental Procedures." The values shown are the means + SD. of triplicate determinations.  on cysteine, experiments were conducted to determine whether mercuric ion could be used to selectively cleave ADPribosyl cysteine linkages in crude extracts containing total liver proteins and, if so, whether ADP-ribose was released intact. For this purpose, transducin modified with radiolabeled ADP-ribose by the action of pertussis toxin was used as a model conjugate. ADP-ribosyl transducin was added to crude extracts of liver proteins and treated with mercuric ion. Fig. IA shows that the radiolabel was released as a function of time in the presence of 10 mM mercuric ion. Approximately 90% of the label was released following a lo-min incubation at 37 "C. Fig. 1B shows the effect of various concentrations of mercuric ion during a lo-min incubation. Approximately 90% of the linkages were released in the presence of 5 mM mercuric ion, and no further release was observed with increasing concentrations. The material released from transducin was examined by strong anion exchange HPLC. Fig. 2A shows a control incubation without mercuric ion, and Fig. 2B shows incubation in the presence of mercuric ion. Released radiolabel co-migrated with an authentic ADP-ribose standard. Crude homogenates of rat liver were subjected to fractionation and analyzed as described under "Experimental Procedures." The results of a representative experiment are shown. The values are the means of triplicate determinations which differed from the mean by less than iO%. a For values shown with the symbol, <, ADP-ribose linkages or enzymatic activity were not detected. The numbers shown represent the limit of detection.
The selectivity of release of ADP-ribose by mercuric ion is shown in Table I. The presence of mercuric ion did not result in cleavage of ADP-ribosyl arginine or ADP-ribosyl diphthamide. The effect of mercuric ion on ADP-ribosyl carboxylates was also examined by analyzing cell extracts (data not shown). The presence of mercuric ion did not release carboxylate ester linkages to ADP-ribose. Further, Aktories et al. (27) have shown that the putative ADP-ribosyl asparagine linkage is also stable in the presence of mercuric ion. Taken together, these results show that, for the known protein ADP-ribose linkages, mercuric ion catalyzes the selective release of ADPribose from cysteine.
Next, total rat liver proteins were examined for the presence of linkages characteristic of ADP-ribosyl cysteine. Trichloroacetic acid-insoluble extracts of liver tissue were treated to remove noncovalently bound ADP-ribose as described previously (28) and subsequently treated with mercuric ion and analyzed for ADP-ribose. Fig. 3 shows such an analysis along with control experiments.
Analysis of an extract treated with mercuric ion revealed a peak that migrated at the expected elution position of etheno-ADP-ribose (Fig. 3A), the fluorescent derivative used for quantifying . Analysis of a parallel sample of liver extract in which mercuric ion was omitted is shown in Fig. 3B. Fig. 3C shows the result obtained when chloroacetaldehyde, which is required for the formation of etheno-ADP-ribose (20), was omitted. This control rules out the possibility that mercuric ion released endogenous fluorescent compounds unrelated to ADP-ribose. Fig. 30 shows that a small amount of authentic etheno-ADP-ribose added to extracts prepared as in Fig. 3A resulted in an enhancement of the peak. Taken together, the results of Fig.  3 demonstrate that rat liver proteins are modified in uiuo with ADP-ribose in linkages chemically indistinguishable from ADP-ribose linked to cysteine.
To further examine endogenous ADP-ribosyl cysteine linkages, crude homogenates of rat liver were subjected to subcellular fractionation using methodology that has been described in detail elsewhere (21-23). The isolated subcellular fractions were analyzed for linkages characteristic of ADP-ribosyl cysteine and total protein. These data along with analysis of an equivalent amount of crude homogenate are shown in Table  II. ADP-ribosyl cysteine linkages were detected in the plasma membrane fraction only. With regard to recovery, this fraction contained 97% of the linkages detected in the crude homogenate. In contrast to the distribution of ADP-ribosyl cysteine linkages, ADP-ribosyl arginine linkages were detected in each of the subcellular fractions except nuclear, with only 4% of the total ADP-ribosyl arginine present in the plasma membrane fraction (results not shown). Na+,K+-ATPase activity was measured as a criteria of purity for the plasma membrane fraction. A specific activity of 10.5 pmol of Pi/mg of protein/h was obtained for this fraction which is in close agreement wtih the value of 11.7 previously reported for a highly purified plasma membrane fraction of rat liver (29). In total, these results indicate that proteins modified on cysteine by ADP-ribose are located primarily, if not exclusively, in the plasma membrane.
The function of endogenous mono-ADP-ribosyltransferases in animal cells is unknown. Since ADP-ribose acceptor proteins have not been identified, these enzymes are presently categorized according to functional group specificity for ADPribose transfer. A number of apparently distinct endogenous transferases that are specific for transferring ADP-ribose to the guanidinium group of free arginine and arginine residues in protein have been characterized (13). The detection of proteins covalently modified by ADP-ribose with linkages indistinguishable from ADP-ribosyl arginine has provided evidence that protein molecules can be in uiuo substrates for these enzymes (17). The presence of proteins that are modified by ADP-ribose with linkages indistinguishable from carboxylate esters of ADP-ribose has also been reported (17). Mono-ADP-ribosyltransferases that catalyze the formation of carboxylate ester linkages to ADP-ribose have not been described. However, such a linkage can be formed by poly(ADPribose) polymerase or the concerted activity of poly(ADPribose) polymerase and poly(ADP-ribose) glycohydrolase (1,30). Thus, the mono-ADP-ribosyl carboxylate ester-like linkages are likely the result of ADP-ribose polymer metabolism.
In the present study, a third class of endogenous ADPribosyl protein linkages has been detected. The linkages are chemically indistinguishable from ADP-ribosyl cysteine. Demonstration of the function of endogenous protein modification by ADP-ribose will require identification of the acceptor proteins. The methods for selective release of ADPribose from protein described here combined with those described previously (17, 20) should prove valuable in this regard. The exclusive location of proteins modified on cysteine in the plasma membrane fraction has interesting implications since pertussis toxin is known to disrupt signal transduction by the modification of a cysteine residue of the plasma membrane protein Gi and related G proteins (6-8). It is of interest that a mono-ADP-ribosyltransferase has been recently detected in human erythrocytes that catalyzes the ADP-ribo-sylation of free cysteine and of Gi,in vitro (16). The results presented here raise the possibility that the modification of protein cysteine residues is involved in the endogenous regulation of signal transduction. Studies on the identity of the specific plasma membrane acceptor proteins modified at cysteine are currently under way.
Post-translational modifications of protein molecules can be subdivided into classes that are stable or metabolically reversible (31). An unanswered question with regard to the modification of proteins by ADP-ribose in animal cells is whether these modifications are reversible. Enzymatic activities have been detected in cultured mouse cells which catalyze the release of ADP-ribose from protein arginine residues (32). This observation supports the possibility that this modification is reversible. Also, the studies of Ludden and co-workers (33-35) have shown that a reversible modification cycle occurs in bacterial cells since nitrogenase of Rhodospirillum rubrum is inactivated and reactivated via cycles of ADP-ribose addition to and removal from a specific arginine residue of the enzyme. It remains to be shown whether or not the modification of proteins at cysteine residues reported here is reversible.